Deep-submicron Complementary Metal-Oxide-Semiconductor (CMOS) is the primary technology for ultra-large scale integrated (ULSI) circuits. Scaling of CMOS has been the principal focus of the microelectronics industry over the last two decades.
As device sizes are scaled down, the source and drain junctions have to be scaled down. As the source and drain are scaled down, series resistance increases, which degrades the device performance. To reduce series resistance, advanced MOS transistors today have relatively deep source and drain junctions to improve series resistance and very shallow source and drain extension junctions to permit the scale down.
With the very shallow source and drain extension junctions which allow the source and drain to be closer together, the channel length between the source and drain is shortened. The shortening in the channel length has lead to several severe problems.
First, with high drive currents, the leakage across the channel is unacceptably high when the transistor is supposed to be off.
Second, the transistors have low manufacturability because of a high range in off current spread between nominal and sub-nominal devices.
Third, the so-called short channel effects (SCE), which degrade the performance of the scaled MOS devices, increase disproportionately with the shortened channel. Basically, the threshold voltage for turning the transistor on and off is nonlinearly lower for sub-0.18 micron transistors than it is for a 0.18 micron transistor and even worse for a sub-0.13 transistor, making it much, much harder to control the smaller transistors.
The conventional approach has been to provide deep sub-micron transistors with laterally uniform channel doping profiles which are scaled down from larger transistors. This approach does not solve the above problems.
Although it has long been known that laterally non-uniform channel doping profiles could be the solution to these problems and although there have been a number of attempts to provide laterally non-uniform channel profiles through direct doping, all of the previous attempts have been uniformly unsuccessful in solving all the problems.